1 /* $FreeBSD: releng/6.3/sys/netinet6/in6_src.c 172047 2007-09-04 22:40:40Z alfred $ */
2 /* $KAME: in6_src.c,v 1.132 2003/08/26 04:42:27 keiichi Exp $ */
3
4 /*-
5 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
6 * All rights reserved.
7 *
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
10 * are met:
11 * 1. Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in the
15 * documentation and/or other materials provided with the distribution.
16 * 3. Neither the name of the project nor the names of its contributors
17 * may be used to endorse or promote products derived from this software
18 * without specific prior written permission.
19 *
20 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30 * SUCH DAMAGE.
31 */
32
33 /*-
34 * Copyright (c) 1982, 1986, 1991, 1993
35 * The Regents of the University of California. All rights reserved.
36 *
37 * Redistribution and use in source and binary forms, with or without
38 * modification, are permitted provided that the following conditions
39 * are met:
40 * 1. Redistributions of source code must retain the above copyright
41 * notice, this list of conditions and the following disclaimer.
42 * 2. Redistributions in binary form must reproduce the above copyright
43 * notice, this list of conditions and the following disclaimer in the
44 * documentation and/or other materials provided with the distribution.
45 * 4. Neither the name of the University nor the names of its contributors
46 * may be used to endorse or promote products derived from this software
47 * without specific prior written permission.
48 *
49 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
50 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
51 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
52 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
53 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
54 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
55 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
56 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
57 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
58 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
59 * SUCH DAMAGE.
60 *
61 * @(#)in_pcb.c 8.2 (Berkeley) 1/4/94
62 */
63
64 #include "opt_inet.h"
65 #include "opt_inet6.h"
66
67 #include <sys/param.h>
68 #include <sys/systm.h>
69 #include <sys/malloc.h>
70 #include <sys/mbuf.h>
71 #include <sys/protosw.h>
72 #include <sys/socket.h>
73 #include <sys/socketvar.h>
74 #include <sys/sockio.h>
75 #include <sys/sysctl.h>
76 #include <sys/errno.h>
77 #include <sys/time.h>
78 #include <sys/kernel.h>
79 #include <sys/lock.h>
80 #include <sys/sx.h>
81
82 #include <net/if.h>
83 #include <net/route.h>
84
85 #include <netinet/in.h>
86 #include <netinet/in_var.h>
87 #include <netinet/in_systm.h>
88 #include <netinet/ip.h>
89 #include <netinet/in_pcb.h>
90 #include <netinet6/in6_var.h>
91 #include <netinet/ip6.h>
92 #include <netinet6/in6_pcb.h>
93 #include <netinet6/ip6_var.h>
94 #include <netinet6/scope6_var.h>
95 #include <netinet6/nd6.h>
96
97 #include <net/net_osdep.h>
98
99 static struct mtx addrsel_lock;
100 #define ADDRSEL_LOCK_INIT() mtx_init(&addrsel_lock, "addrsel_lock", NULL, MTX_DEF)
101 #define ADDRSEL_LOCK() mtx_lock(&addrsel_lock)
102 #define ADDRSEL_UNLOCK() mtx_unlock(&addrsel_lock)
103 #define ADDRSEL_LOCK_ASSERT() mtx_assert(&addrsel_lock, MA_OWNED)
104
105 static struct sx addrsel_sxlock;
106 #define ADDRSEL_SXLOCK_INIT() sx_init(&addrsel_sxlock, "addrsel_sxlock")
107 #define ADDRSEL_SLOCK() sx_slock(&addrsel_sxlock)
108 #define ADDRSEL_SUNLOCK() sx_sunlock(&addrsel_sxlock)
109 #define ADDRSEL_XLOCK() sx_xlock(&addrsel_sxlock)
110 #define ADDRSEL_XUNLOCK() sx_xunlock(&addrsel_sxlock)
111
112 #define ADDR_LABEL_NOTAPP (-1)
113 struct in6_addrpolicy defaultaddrpolicy;
114
115 int ip6_prefer_tempaddr = 0;
116
117 static int selectroute __P((struct sockaddr_in6 *, struct ip6_pktopts *,
118 struct ip6_moptions *, struct route_in6 *, struct ifnet **,
119 struct rtentry **, int, int));
120 static int in6_selectif __P((struct sockaddr_in6 *, struct ip6_pktopts *,
121 struct ip6_moptions *, struct route_in6 *ro, struct ifnet **));
122
123 static struct in6_addrpolicy *lookup_addrsel_policy __P((struct sockaddr_in6 *));
124
125 static void init_policy_queue __P((void));
126 static int add_addrsel_policyent __P((struct in6_addrpolicy *));
127 static int delete_addrsel_policyent __P((struct in6_addrpolicy *));
128 static int walk_addrsel_policy __P((int (*)(struct in6_addrpolicy *, void *),
129 void *));
130 static int dump_addrsel_policyent __P((struct in6_addrpolicy *, void *));
131 static struct in6_addrpolicy *match_addrsel_policy __P((struct sockaddr_in6 *));
132
133 /*
134 * Return an IPv6 address, which is the most appropriate for a given
135 * destination and user specified options.
136 * If necessary, this function lookups the routing table and returns
137 * an entry to the caller for later use.
138 */
139 #define REPLACE(r) do {\
140 if ((r) < sizeof(ip6stat.ip6s_sources_rule) / \
141 sizeof(ip6stat.ip6s_sources_rule[0])) /* check for safety */ \
142 ip6stat.ip6s_sources_rule[(r)]++; \
143 /* printf("in6_selectsrc: replace %s with %s by %d\n", ia_best ? ip6_sprintf(&ia_best->ia_addr.sin6_addr) : "none", ip6_sprintf(&ia->ia_addr.sin6_addr), (r)); */ \
144 goto replace; \
145 } while(0)
146 #define NEXT(r) do {\
147 if ((r) < sizeof(ip6stat.ip6s_sources_rule) / \
148 sizeof(ip6stat.ip6s_sources_rule[0])) /* check for safety */ \
149 ip6stat.ip6s_sources_rule[(r)]++; \
150 /* printf("in6_selectsrc: keep %s against %s by %d\n", ia_best ? ip6_sprintf(&ia_best->ia_addr.sin6_addr) : "none", ip6_sprintf(&ia->ia_addr.sin6_addr), (r)); */ \
151 goto next; /* XXX: we can't use 'continue' here */ \
152 } while(0)
153 #define BREAK(r) do { \
154 if ((r) < sizeof(ip6stat.ip6s_sources_rule) / \
155 sizeof(ip6stat.ip6s_sources_rule[0])) /* check for safety */ \
156 ip6stat.ip6s_sources_rule[(r)]++; \
157 goto out; /* XXX: we can't use 'break' here */ \
158 } while(0)
159
160 struct in6_addr *
161 in6_selectsrc(dstsock, opts, mopts, ro, laddr, ifpp, errorp)
162 struct sockaddr_in6 *dstsock;
163 struct ip6_pktopts *opts;
164 struct ip6_moptions *mopts;
165 struct route_in6 *ro;
166 struct in6_addr *laddr;
167 struct ifnet **ifpp;
168 int *errorp;
169 {
170 struct in6_addr dst;
171 struct ifnet *ifp = NULL;
172 struct in6_ifaddr *ia = NULL, *ia_best = NULL;
173 struct in6_pktinfo *pi = NULL;
174 int dst_scope = -1, best_scope = -1, best_matchlen = -1;
175 struct in6_addrpolicy *dst_policy = NULL, *best_policy = NULL;
176 u_int32_t odstzone;
177 int prefer_tempaddr;
178
179 dst = dstsock->sin6_addr; /* make a copy for local operation */
180 *errorp = 0;
181 if (ifpp)
182 *ifpp = NULL;
183
184 /*
185 * If the source address is explicitly specified by the caller,
186 * check if the requested source address is indeed a unicast address
187 * assigned to the node, and can be used as the packet's source
188 * address. If everything is okay, use the address as source.
189 */
190 if (opts && (pi = opts->ip6po_pktinfo) &&
191 !IN6_IS_ADDR_UNSPECIFIED(&pi->ipi6_addr)) {
192 struct sockaddr_in6 srcsock;
193 struct in6_ifaddr *ia6;
194
195 /* get the outgoing interface */
196 if ((*errorp = in6_selectif(dstsock, opts, mopts, ro, &ifp))
197 != 0) {
198 return (NULL);
199 }
200
201 /*
202 * determine the appropriate zone id of the source based on
203 * the zone of the destination and the outgoing interface.
204 * If the specified address is ambiguous wrt the scope zone,
205 * the interface must be specified; otherwise, ifa_ifwithaddr()
206 * will fail matching the address.
207 */
208 bzero(&srcsock, sizeof(srcsock));
209 srcsock.sin6_family = AF_INET6;
210 srcsock.sin6_len = sizeof(srcsock);
211 srcsock.sin6_addr = pi->ipi6_addr;
212 if (ifp) {
213 *errorp = in6_setscope(&srcsock.sin6_addr, ifp, NULL);
214 if (*errorp != 0)
215 return (NULL);
216 }
217
218 ia6 = (struct in6_ifaddr *)ifa_ifwithaddr((struct sockaddr *)(&srcsock));
219 if (ia6 == NULL ||
220 (ia6->ia6_flags & (IN6_IFF_ANYCAST | IN6_IFF_NOTREADY))) {
221 *errorp = EADDRNOTAVAIL;
222 return (NULL);
223 }
224 pi->ipi6_addr = srcsock.sin6_addr; /* XXX: this overrides pi */
225 if (ifpp)
226 *ifpp = ifp;
227 return (&ia6->ia_addr.sin6_addr);
228 }
229
230 /*
231 * Otherwise, if the socket has already bound the source, just use it.
232 */
233 if (laddr && !IN6_IS_ADDR_UNSPECIFIED(laddr))
234 return (laddr);
235
236 /*
237 * If the address is not specified, choose the best one based on
238 * the outgoing interface and the destination address.
239 */
240 /* get the outgoing interface */
241 if ((*errorp = in6_selectif(dstsock, opts, mopts, ro, &ifp)) != 0)
242 return (NULL);
243
244 #ifdef DIAGNOSTIC
245 if (ifp == NULL) /* this should not happen */
246 panic("in6_selectsrc: NULL ifp");
247 #endif
248 *errorp = in6_setscope(&dst, ifp, &odstzone);
249 if (*errorp != 0)
250 return (NULL);
251
252 for (ia = in6_ifaddr; ia; ia = ia->ia_next) {
253 int new_scope = -1, new_matchlen = -1;
254 struct in6_addrpolicy *new_policy = NULL;
255 u_int32_t srczone, osrczone, dstzone;
256 struct in6_addr src;
257 struct ifnet *ifp1 = ia->ia_ifp;
258
259 /*
260 * We'll never take an address that breaks the scope zone
261 * of the destination. We also skip an address if its zone
262 * does not contain the outgoing interface.
263 * XXX: we should probably use sin6_scope_id here.
264 */
265 if (in6_setscope(&dst, ifp1, &dstzone) ||
266 odstzone != dstzone) {
267 continue;
268 }
269 src = ia->ia_addr.sin6_addr;
270 if (in6_setscope(&src, ifp, &osrczone) ||
271 in6_setscope(&src, ifp1, &srczone) ||
272 osrczone != srczone) {
273 continue;
274 }
275
276 /* avoid unusable addresses */
277 if ((ia->ia6_flags &
278 (IN6_IFF_NOTREADY | IN6_IFF_ANYCAST | IN6_IFF_DETACHED))) {
279 continue;
280 }
281 if (!ip6_use_deprecated && IFA6_IS_DEPRECATED(ia))
282 continue;
283
284 /* Rule 1: Prefer same address */
285 if (IN6_ARE_ADDR_EQUAL(&dst, &ia->ia_addr.sin6_addr)) {
286 ia_best = ia;
287 BREAK(1); /* there should be no better candidate */
288 }
289
290 if (ia_best == NULL)
291 REPLACE(0);
292
293 /* Rule 2: Prefer appropriate scope */
294 if (dst_scope < 0)
295 dst_scope = in6_addrscope(&dst);
296 new_scope = in6_addrscope(&ia->ia_addr.sin6_addr);
297 if (IN6_ARE_SCOPE_CMP(best_scope, new_scope) < 0) {
298 if (IN6_ARE_SCOPE_CMP(best_scope, dst_scope) < 0)
299 REPLACE(2);
300 NEXT(2);
301 } else if (IN6_ARE_SCOPE_CMP(new_scope, best_scope) < 0) {
302 if (IN6_ARE_SCOPE_CMP(new_scope, dst_scope) < 0)
303 NEXT(2);
304 REPLACE(2);
305 }
306
307 /*
308 * Rule 3: Avoid deprecated addresses. Note that the case of
309 * !ip6_use_deprecated is already rejected above.
310 */
311 if (!IFA6_IS_DEPRECATED(ia_best) && IFA6_IS_DEPRECATED(ia))
312 NEXT(3);
313 if (IFA6_IS_DEPRECATED(ia_best) && !IFA6_IS_DEPRECATED(ia))
314 REPLACE(3);
315
316 /* Rule 4: Prefer home addresses */
317 /*
318 * XXX: This is a TODO. We should probably merge the MIP6
319 * case above.
320 */
321
322 /* Rule 5: Prefer outgoing interface */
323 if (ia_best->ia_ifp == ifp && ia->ia_ifp != ifp)
324 NEXT(5);
325 if (ia_best->ia_ifp != ifp && ia->ia_ifp == ifp)
326 REPLACE(5);
327
328 /*
329 * Rule 6: Prefer matching label
330 * Note that best_policy should be non-NULL here.
331 */
332 if (dst_policy == NULL)
333 dst_policy = lookup_addrsel_policy(dstsock);
334 if (dst_policy->label != ADDR_LABEL_NOTAPP) {
335 new_policy = lookup_addrsel_policy(&ia->ia_addr);
336 if (dst_policy->label == best_policy->label &&
337 dst_policy->label != new_policy->label)
338 NEXT(6);
339 if (dst_policy->label != best_policy->label &&
340 dst_policy->label == new_policy->label)
341 REPLACE(6);
342 }
343
344 /*
345 * Rule 7: Prefer public addresses.
346 * We allow users to reverse the logic by configuring
347 * a sysctl variable, so that privacy conscious users can
348 * always prefer temporary addresses.
349 */
350 if (opts == NULL ||
351 opts->ip6po_prefer_tempaddr == IP6PO_TEMPADDR_SYSTEM) {
352 prefer_tempaddr = ip6_prefer_tempaddr;
353 } else if (opts->ip6po_prefer_tempaddr ==
354 IP6PO_TEMPADDR_NOTPREFER) {
355 prefer_tempaddr = 0;
356 } else
357 prefer_tempaddr = 1;
358 if (!(ia_best->ia6_flags & IN6_IFF_TEMPORARY) &&
359 (ia->ia6_flags & IN6_IFF_TEMPORARY)) {
360 if (prefer_tempaddr)
361 REPLACE(7);
362 else
363 NEXT(7);
364 }
365 if ((ia_best->ia6_flags & IN6_IFF_TEMPORARY) &&
366 !(ia->ia6_flags & IN6_IFF_TEMPORARY)) {
367 if (prefer_tempaddr)
368 NEXT(7);
369 else
370 REPLACE(7);
371 }
372
373 /*
374 * Rule 8: prefer addresses on alive interfaces.
375 * This is a KAME specific rule.
376 */
377 if ((ia_best->ia_ifp->if_flags & IFF_UP) &&
378 !(ia->ia_ifp->if_flags & IFF_UP))
379 NEXT(8);
380 if (!(ia_best->ia_ifp->if_flags & IFF_UP) &&
381 (ia->ia_ifp->if_flags & IFF_UP))
382 REPLACE(8);
383
384 /*
385 * Rule 14: Use longest matching prefix.
386 * Note: in the address selection draft, this rule is
387 * documented as "Rule 8". However, since it is also
388 * documented that this rule can be overridden, we assign
389 * a large number so that it is easy to assign smaller numbers
390 * to more preferred rules.
391 */
392 new_matchlen = in6_matchlen(&ia->ia_addr.sin6_addr, &dst);
393 if (best_matchlen < new_matchlen)
394 REPLACE(14);
395 if (new_matchlen < best_matchlen)
396 NEXT(14);
397
398 /* Rule 15 is reserved. */
399
400 /*
401 * Last resort: just keep the current candidate.
402 * Or, do we need more rules?
403 */
404 continue;
405
406 replace:
407 ia_best = ia;
408 best_scope = (new_scope >= 0 ? new_scope :
409 in6_addrscope(&ia_best->ia_addr.sin6_addr));
410 best_policy = (new_policy ? new_policy :
411 lookup_addrsel_policy(&ia_best->ia_addr));
412 best_matchlen = (new_matchlen >= 0 ? new_matchlen :
413 in6_matchlen(&ia_best->ia_addr.sin6_addr,
414 &dst));
415
416 next:
417 continue;
418
419 out:
420 break;
421 }
422
423 if ((ia = ia_best) == NULL) {
424 *errorp = EADDRNOTAVAIL;
425 return (NULL);
426 }
427
428 if (ifpp)
429 *ifpp = ifp;
430
431 return (&ia->ia_addr.sin6_addr);
432 }
433
434 static int
435 selectroute(dstsock, opts, mopts, ro, retifp, retrt, clone, norouteok)
436 struct sockaddr_in6 *dstsock;
437 struct ip6_pktopts *opts;
438 struct ip6_moptions *mopts;
439 struct route_in6 *ro;
440 struct ifnet **retifp;
441 struct rtentry **retrt;
442 int clone; /* meaningful only for bsdi and freebsd. */
443 int norouteok;
444 {
445 int error = 0;
446 struct ifnet *ifp = NULL;
447 struct rtentry *rt = NULL;
448 struct sockaddr_in6 *sin6_next;
449 struct in6_pktinfo *pi = NULL;
450 struct in6_addr *dst = &dstsock->sin6_addr;
451
452 #if 0
453 if (dstsock->sin6_addr.s6_addr32[0] == 0 &&
454 dstsock->sin6_addr.s6_addr32[1] == 0 &&
455 !IN6_IS_ADDR_LOOPBACK(&dstsock->sin6_addr)) {
456 printf("in6_selectroute: strange destination %s\n",
457 ip6_sprintf(&dstsock->sin6_addr));
458 } else {
459 printf("in6_selectroute: destination = %s%%%d\n",
460 ip6_sprintf(&dstsock->sin6_addr),
461 dstsock->sin6_scope_id); /* for debug */
462 }
463 #endif
464
465 /* If the caller specify the outgoing interface explicitly, use it. */
466 if (opts && (pi = opts->ip6po_pktinfo) != NULL && pi->ipi6_ifindex) {
467 /* XXX boundary check is assumed to be already done. */
468 ifp = ifnet_byindex(pi->ipi6_ifindex);
469 if (ifp != NULL &&
470 (norouteok || retrt == NULL ||
471 IN6_IS_ADDR_MULTICAST(dst))) {
472 /*
473 * we do not have to check or get the route for
474 * multicast.
475 */
476 goto done;
477 } else
478 goto getroute;
479 }
480
481 /*
482 * If the destination address is a multicast address and the outgoing
483 * interface for the address is specified by the caller, use it.
484 */
485 if (IN6_IS_ADDR_MULTICAST(dst) &&
486 mopts != NULL && (ifp = mopts->im6o_multicast_ifp) != NULL) {
487 goto done; /* we do not need a route for multicast. */
488 }
489
490 getroute:
491 /*
492 * If the next hop address for the packet is specified by the caller,
493 * use it as the gateway.
494 */
495 if (opts && opts->ip6po_nexthop) {
496 struct route_in6 *ron;
497
498 sin6_next = satosin6(opts->ip6po_nexthop);
499
500 /* at this moment, we only support AF_INET6 next hops */
501 if (sin6_next->sin6_family != AF_INET6) {
502 error = EAFNOSUPPORT; /* or should we proceed? */
503 goto done;
504 }
505
506 /*
507 * If the next hop is an IPv6 address, then the node identified
508 * by that address must be a neighbor of the sending host.
509 */
510 ron = &opts->ip6po_nextroute;
511 if ((ron->ro_rt &&
512 (ron->ro_rt->rt_flags & (RTF_UP | RTF_LLINFO)) !=
513 (RTF_UP | RTF_LLINFO)) ||
514 !IN6_ARE_ADDR_EQUAL(&satosin6(&ron->ro_dst)->sin6_addr,
515 &sin6_next->sin6_addr)) {
516 if (ron->ro_rt) {
517 RTFREE(ron->ro_rt);
518 ron->ro_rt = NULL;
519 }
520 *satosin6(&ron->ro_dst) = *sin6_next;
521 }
522 if (ron->ro_rt == NULL) {
523 rtalloc((struct route *)ron); /* multi path case? */
524 if (ron->ro_rt == NULL ||
525 !(ron->ro_rt->rt_flags & RTF_LLINFO)) {
526 if (ron->ro_rt) {
527 RTFREE(ron->ro_rt);
528 ron->ro_rt = NULL;
529 }
530 error = EHOSTUNREACH;
531 goto done;
532 }
533 }
534 rt = ron->ro_rt;
535 ifp = rt->rt_ifp;
536
537 /*
538 * When cloning is required, try to allocate a route to the
539 * destination so that the caller can store path MTU
540 * information.
541 */
542 if (!clone)
543 goto done;
544 }
545
546 /*
547 * Use a cached route if it exists and is valid, else try to allocate
548 * a new one. Note that we should check the address family of the
549 * cached destination, in case of sharing the cache with IPv4.
550 */
551 if (ro) {
552 if (ro->ro_rt &&
553 (!(ro->ro_rt->rt_flags & RTF_UP) ||
554 ((struct sockaddr *)(&ro->ro_dst))->sa_family != AF_INET6 ||
555 !IN6_ARE_ADDR_EQUAL(&satosin6(&ro->ro_dst)->sin6_addr,
556 dst))) {
557 RTFREE(ro->ro_rt);
558 ro->ro_rt = (struct rtentry *)NULL;
559 }
560 if (ro->ro_rt == (struct rtentry *)NULL) {
561 struct sockaddr_in6 *sa6;
562
563 /* No route yet, so try to acquire one */
564 bzero(&ro->ro_dst, sizeof(struct sockaddr_in6));
565 sa6 = (struct sockaddr_in6 *)&ro->ro_dst;
566 *sa6 = *dstsock;
567 sa6->sin6_scope_id = 0;
568
569 if (clone) {
570 rtalloc((struct route *)ro);
571 } else {
572 ro->ro_rt = rtalloc1(&((struct route *)ro)
573 ->ro_dst, 0, 0UL);
574 if (ro->ro_rt)
575 RT_UNLOCK(ro->ro_rt);
576 }
577 }
578
579 /*
580 * do not care about the result if we have the nexthop
581 * explicitly specified.
582 */
583 if (opts && opts->ip6po_nexthop)
584 goto done;
585
586 if (ro->ro_rt) {
587 ifp = ro->ro_rt->rt_ifp;
588
589 if (ifp == NULL) { /* can this really happen? */
590 RTFREE(ro->ro_rt);
591 ro->ro_rt = NULL;
592 }
593 }
594 if (ro->ro_rt == NULL)
595 error = EHOSTUNREACH;
596 rt = ro->ro_rt;
597
598 /*
599 * Check if the outgoing interface conflicts with
600 * the interface specified by ipi6_ifindex (if specified).
601 * Note that loopback interface is always okay.
602 * (this may happen when we are sending a packet to one of
603 * our own addresses.)
604 */
605 if (ifp && opts && opts->ip6po_pktinfo &&
606 opts->ip6po_pktinfo->ipi6_ifindex) {
607 if (!(ifp->if_flags & IFF_LOOPBACK) &&
608 ifp->if_index !=
609 opts->ip6po_pktinfo->ipi6_ifindex) {
610 error = EHOSTUNREACH;
611 goto done;
612 }
613 }
614 }
615
616 done:
617 if (ifp == NULL && rt == NULL) {
618 /*
619 * This can happen if the caller did not pass a cached route
620 * nor any other hints. We treat this case an error.
621 */
622 error = EHOSTUNREACH;
623 }
624 if (error == EHOSTUNREACH)
625 ip6stat.ip6s_noroute++;
626
627 if (retifp != NULL)
628 *retifp = ifp;
629 if (retrt != NULL)
630 *retrt = rt; /* rt may be NULL */
631
632 return (error);
633 }
634
635 static int
636 in6_selectif(dstsock, opts, mopts, ro, retifp)
637 struct sockaddr_in6 *dstsock;
638 struct ip6_pktopts *opts;
639 struct ip6_moptions *mopts;
640 struct route_in6 *ro;
641 struct ifnet **retifp;
642 {
643 int error;
644 struct route_in6 sro;
645 struct rtentry *rt = NULL;
646
647 if (ro == NULL) {
648 bzero(&sro, sizeof(sro));
649 ro = &sro;
650 }
651
652 if ((error = selectroute(dstsock, opts, mopts, ro, retifp,
653 &rt, 0, 1)) != 0) {
654 if (rt && rt == sro.ro_rt)
655 RTFREE(rt);
656 return (error);
657 }
658
659 /*
660 * do not use a rejected or black hole route.
661 * XXX: this check should be done in the L2 output routine.
662 * However, if we skipped this check here, we'd see the following
663 * scenario:
664 * - install a rejected route for a scoped address prefix
665 * (like fe80::/10)
666 * - send a packet to a destination that matches the scoped prefix,
667 * with ambiguity about the scope zone.
668 * - pick the outgoing interface from the route, and disambiguate the
669 * scope zone with the interface.
670 * - ip6_output() would try to get another route with the "new"
671 * destination, which may be valid.
672 * - we'd see no error on output.
673 * Although this may not be very harmful, it should still be confusing.
674 * We thus reject the case here.
675 */
676 if (rt && (rt->rt_flags & (RTF_REJECT | RTF_BLACKHOLE))) {
677 int flags = (rt->rt_flags & RTF_HOST ? EHOSTUNREACH : ENETUNREACH);
678
679 if (rt && rt == sro.ro_rt)
680 RTFREE(rt);
681 return (flags);
682 }
683
684 /*
685 * Adjust the "outgoing" interface. If we're going to loop the packet
686 * back to ourselves, the ifp would be the loopback interface.
687 * However, we'd rather know the interface associated to the
688 * destination address (which should probably be one of our own
689 * addresses.)
690 */
691 if (rt && rt->rt_ifa && rt->rt_ifa->ifa_ifp)
692 *retifp = rt->rt_ifa->ifa_ifp;
693
694 if (rt && rt == sro.ro_rt)
695 RTFREE(rt);
696 return (0);
697 }
698
699 int
700 in6_selectroute(dstsock, opts, mopts, ro, retifp, retrt, clone)
701 struct sockaddr_in6 *dstsock;
702 struct ip6_pktopts *opts;
703 struct ip6_moptions *mopts;
704 struct route_in6 *ro;
705 struct ifnet **retifp;
706 struct rtentry **retrt;
707 int clone; /* meaningful only for bsdi and freebsd. */
708 {
709 return (selectroute(dstsock, opts, mopts, ro, retifp,
710 retrt, clone, 0));
711 }
712
713 /*
714 * Default hop limit selection. The precedence is as follows:
715 * 1. Hoplimit value specified via ioctl.
716 * 2. (If the outgoing interface is detected) the current
717 * hop limit of the interface specified by router advertisement.
718 * 3. The system default hoplimit.
719 */
720 int
721 in6_selecthlim(in6p, ifp)
722 struct in6pcb *in6p;
723 struct ifnet *ifp;
724 {
725 if (in6p && in6p->in6p_hops >= 0)
726 return (in6p->in6p_hops);
727 else if (ifp)
728 return (ND_IFINFO(ifp)->chlim);
729 else if (in6p && !IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr)) {
730 struct route_in6 ro6;
731 struct ifnet *lifp;
732
733 bzero(&ro6, sizeof(ro6));
734 ro6.ro_dst.sin6_family = AF_INET6;
735 ro6.ro_dst.sin6_len = sizeof(struct sockaddr_in6);
736 ro6.ro_dst.sin6_addr = in6p->in6p_faddr;
737 rtalloc((struct route *)&ro6);
738 if (ro6.ro_rt) {
739 lifp = ro6.ro_rt->rt_ifp;
740 RTFREE(ro6.ro_rt);
741 if (lifp)
742 return (ND_IFINFO(lifp)->chlim);
743 } else
744 return (ip6_defhlim);
745 }
746 return (ip6_defhlim);
747 }
748
749 /*
750 * XXX: this is borrowed from in6_pcbbind(). If possible, we should
751 * share this function by all *bsd*...
752 */
753 int
754 in6_pcbsetport(laddr, inp, cred)
755 struct in6_addr *laddr;
756 struct inpcb *inp;
757 struct ucred *cred;
758 {
759 struct socket *so = inp->inp_socket;
760 u_int16_t lport = 0, first, last, *lastport;
761 int count, error = 0, wild = 0;
762 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
763
764 /* XXX: this is redundant when called from in6_pcbbind */
765 if ((so->so_options & (SO_REUSEADDR|SO_REUSEPORT)) == 0)
766 wild = INPLOOKUP_WILDCARD;
767
768 inp->inp_flags |= INP_ANONPORT;
769
770 if (inp->inp_flags & INP_HIGHPORT) {
771 first = ipport_hifirstauto; /* sysctl */
772 last = ipport_hilastauto;
773 lastport = &pcbinfo->lasthi;
774 } else if (inp->inp_flags & INP_LOWPORT) {
775 if ((error = suser_cred(cred, 0)))
776 return error;
777 first = ipport_lowfirstauto; /* 1023 */
778 last = ipport_lowlastauto; /* 600 */
779 lastport = &pcbinfo->lastlow;
780 } else {
781 first = ipport_firstauto; /* sysctl */
782 last = ipport_lastauto;
783 lastport = &pcbinfo->lastport;
784 }
785 /*
786 * Simple check to ensure all ports are not used up causing
787 * a deadlock here.
788 *
789 * We split the two cases (up and down) so that the direction
790 * is not being tested on each round of the loop.
791 */
792 if (first > last) {
793 /*
794 * counting down
795 */
796 count = first - last;
797
798 do {
799 if (count-- < 0) { /* completely used? */
800 /*
801 * Undo any address bind that may have
802 * occurred above.
803 */
804 inp->in6p_laddr = in6addr_any;
805 return (EAGAIN);
806 }
807 --*lastport;
808 if (*lastport > first || *lastport < last)
809 *lastport = first;
810 lport = htons(*lastport);
811 } while (in6_pcblookup_local(pcbinfo, &inp->in6p_laddr,
812 lport, wild));
813 } else {
814 /*
815 * counting up
816 */
817 count = last - first;
818
819 do {
820 if (count-- < 0) { /* completely used? */
821 /*
822 * Undo any address bind that may have
823 * occurred above.
824 */
825 inp->in6p_laddr = in6addr_any;
826 return (EAGAIN);
827 }
828 ++*lastport;
829 if (*lastport < first || *lastport > last)
830 *lastport = first;
831 lport = htons(*lastport);
832 } while (in6_pcblookup_local(pcbinfo,
833 &inp->in6p_laddr, lport, wild));
834 }
835
836 inp->inp_lport = lport;
837 if (in_pcbinshash(inp) != 0) {
838 inp->in6p_laddr = in6addr_any;
839 inp->inp_lport = 0;
840 return (EAGAIN);
841 }
842
843 return (0);
844 }
845
846 void
847 addrsel_policy_init()
848 {
849 ADDRSEL_LOCK_INIT();
850 ADDRSEL_SXLOCK_INIT();
851
852 init_policy_queue();
853
854 /* initialize the "last resort" policy */
855 bzero(&defaultaddrpolicy, sizeof(defaultaddrpolicy));
856 defaultaddrpolicy.label = ADDR_LABEL_NOTAPP;
857 }
858
859 static struct in6_addrpolicy *
860 lookup_addrsel_policy(key)
861 struct sockaddr_in6 *key;
862 {
863 struct in6_addrpolicy *match = NULL;
864
865 ADDRSEL_LOCK();
866 match = match_addrsel_policy(key);
867
868 if (match == NULL)
869 match = &defaultaddrpolicy;
870 else
871 match->use++;
872 ADDRSEL_UNLOCK();
873
874 return (match);
875 }
876
877 /*
878 * Subroutines to manage the address selection policy table via sysctl.
879 */
880 struct walkarg {
881 struct sysctl_req *w_req;
882 };
883
884 static int in6_src_sysctl(SYSCTL_HANDLER_ARGS);
885 SYSCTL_DECL(_net_inet6_ip6);
886 SYSCTL_NODE(_net_inet6_ip6, IPV6CTL_ADDRCTLPOLICY, addrctlpolicy,
887 CTLFLAG_RD, in6_src_sysctl, "");
888
889 static int
890 in6_src_sysctl(SYSCTL_HANDLER_ARGS)
891 {
892 struct walkarg w;
893
894 if (req->newptr)
895 return EPERM;
896
897 bzero(&w, sizeof(w));
898 w.w_req = req;
899
900 return (walk_addrsel_policy(dump_addrsel_policyent, &w));
901 }
902
903 int
904 in6_src_ioctl(cmd, data)
905 u_long cmd;
906 caddr_t data;
907 {
908 int i;
909 struct in6_addrpolicy ent0;
910
911 if (cmd != SIOCAADDRCTL_POLICY && cmd != SIOCDADDRCTL_POLICY)
912 return (EOPNOTSUPP); /* check for safety */
913
914 ent0 = *(struct in6_addrpolicy *)data;
915
916 if (ent0.label == ADDR_LABEL_NOTAPP)
917 return (EINVAL);
918 /* check if the prefix mask is consecutive. */
919 if (in6_mask2len(&ent0.addrmask.sin6_addr, NULL) < 0)
920 return (EINVAL);
921 /* clear trailing garbages (if any) of the prefix address. */
922 for (i = 0; i < 4; i++) {
923 ent0.addr.sin6_addr.s6_addr32[i] &=
924 ent0.addrmask.sin6_addr.s6_addr32[i];
925 }
926 ent0.use = 0;
927
928 switch (cmd) {
929 case SIOCAADDRCTL_POLICY:
930 return (add_addrsel_policyent(&ent0));
931 case SIOCDADDRCTL_POLICY:
932 return (delete_addrsel_policyent(&ent0));
933 }
934
935 return (0); /* XXX: compromise compilers */
936 }
937
938 /*
939 * The followings are implementation of the policy table using a
940 * simple tail queue.
941 * XXX such details should be hidden.
942 * XXX implementation using binary tree should be more efficient.
943 */
944 struct addrsel_policyent {
945 TAILQ_ENTRY(addrsel_policyent) ape_entry;
946 struct in6_addrpolicy ape_policy;
947 };
948
949 TAILQ_HEAD(addrsel_policyhead, addrsel_policyent);
950
951 struct addrsel_policyhead addrsel_policytab;
952
953 static void
954 init_policy_queue()
955 {
956 TAILQ_INIT(&addrsel_policytab);
957 }
958
959 static int
960 add_addrsel_policyent(newpolicy)
961 struct in6_addrpolicy *newpolicy;
962 {
963 struct addrsel_policyent *new, *pol;
964
965 MALLOC(new, struct addrsel_policyent *, sizeof(*new), M_IFADDR,
966 M_WAITOK);
967 ADDRSEL_XLOCK();
968 ADDRSEL_LOCK();
969
970 /* duplication check */
971 TAILQ_FOREACH(pol, &addrsel_policytab, ape_entry) {
972 if (IN6_ARE_ADDR_EQUAL(&newpolicy->addr.sin6_addr,
973 &pol->ape_policy.addr.sin6_addr) &&
974 IN6_ARE_ADDR_EQUAL(&newpolicy->addrmask.sin6_addr,
975 &pol->ape_policy.addrmask.sin6_addr)) {
976 ADDRSEL_UNLOCK();
977 ADDRSEL_XUNLOCK();
978 FREE(new, M_IFADDR);
979 return (EEXIST); /* or override it? */
980 }
981 }
982
983 bzero(new, sizeof(*new));
984
985 /* XXX: should validate entry */
986 new->ape_policy = *newpolicy;
987
988 TAILQ_INSERT_TAIL(&addrsel_policytab, new, ape_entry);
989 ADDRSEL_UNLOCK();
990 ADDRSEL_XUNLOCK();
991
992 return (0);
993 }
994
995 static int
996 delete_addrsel_policyent(key)
997 struct in6_addrpolicy *key;
998 {
999 struct addrsel_policyent *pol;
1000
1001 ADDRSEL_XLOCK();
1002 ADDRSEL_LOCK();
1003
1004 /* search for the entry in the table */
1005 TAILQ_FOREACH(pol, &addrsel_policytab, ape_entry) {
1006 if (IN6_ARE_ADDR_EQUAL(&key->addr.sin6_addr,
1007 &pol->ape_policy.addr.sin6_addr) &&
1008 IN6_ARE_ADDR_EQUAL(&key->addrmask.sin6_addr,
1009 &pol->ape_policy.addrmask.sin6_addr)) {
1010 break;
1011 }
1012 }
1013 if (pol == NULL) {
1014 ADDRSEL_UNLOCK();
1015 ADDRSEL_XUNLOCK();
1016 return (ESRCH);
1017 }
1018
1019 TAILQ_REMOVE(&addrsel_policytab, pol, ape_entry);
1020 ADDRSEL_UNLOCK();
1021 ADDRSEL_XUNLOCK();
1022
1023 return (0);
1024 }
1025
1026 static int
1027 walk_addrsel_policy(callback, w)
1028 int (*callback) __P((struct in6_addrpolicy *, void *));
1029 void *w;
1030 {
1031 struct addrsel_policyent *pol;
1032 int error = 0;
1033
1034 ADDRSEL_SLOCK();
1035 TAILQ_FOREACH(pol, &addrsel_policytab, ape_entry) {
1036 if ((error = (*callback)(&pol->ape_policy, w)) != 0) {
1037 ADDRSEL_SUNLOCK();
1038 return (error);
1039 }
1040 }
1041 ADDRSEL_SUNLOCK();
1042 return (error);
1043 }
1044
1045 static int
1046 dump_addrsel_policyent(pol, arg)
1047 struct in6_addrpolicy *pol;
1048 void *arg;
1049 {
1050 int error = 0;
1051 struct walkarg *w = arg;
1052
1053 error = SYSCTL_OUT(w->w_req, pol, sizeof(*pol));
1054
1055 return (error);
1056 }
1057
1058 static struct in6_addrpolicy *
1059 match_addrsel_policy(key)
1060 struct sockaddr_in6 *key;
1061 {
1062 struct addrsel_policyent *pent;
1063 struct in6_addrpolicy *bestpol = NULL, *pol;
1064 int matchlen, bestmatchlen = -1;
1065 u_char *mp, *ep, *k, *p, m;
1066
1067 TAILQ_FOREACH(pent, &addrsel_policytab, ape_entry) {
1068 matchlen = 0;
1069
1070 pol = &pent->ape_policy;
1071 mp = (u_char *)&pol->addrmask.sin6_addr;
1072 ep = mp + 16; /* XXX: scope field? */
1073 k = (u_char *)&key->sin6_addr;
1074 p = (u_char *)&pol->addr.sin6_addr;
1075 for (; mp < ep && *mp; mp++, k++, p++) {
1076 m = *mp;
1077 if ((*k & m) != *p)
1078 goto next; /* not match */
1079 if (m == 0xff) /* short cut for a typical case */
1080 matchlen += 8;
1081 else {
1082 while (m >= 0x80) {
1083 matchlen++;
1084 m <<= 1;
1085 }
1086 }
1087 }
1088
1089 /* matched. check if this is better than the current best. */
1090 if (bestpol == NULL ||
1091 matchlen > bestmatchlen) {
1092 bestpol = pol;
1093 bestmatchlen = matchlen;
1094 }
1095
1096 next:
1097 continue;
1098 }
1099
1100 return (bestpol);
1101 }
Cache object: 4557fd154a97fbc745f395be51308290
|